Author: Robert Hyatt
Date: 06:56:09 09/16/99
Go up one level in this thread
On September 16, 1999 at 00:57:28, James B. Shearer wrote: >On September 14, 1999 at 10:06:01, Robert Hyatt wrote: > >>On September 14, 1999 at 03:05:35, James B. Shearer wrote: >> >>>On September 13, 1999 at 22:11:25, Robert Hyatt wrote: >>> >>>>If you aren't an IM, your chances of winning a blitz game are essentially zero. >>>>At least a real crafty, on real hardware, with reasonable book, and not >>>>intentionally weakened... It is _very_ difficult for IM players to beat it, >>>>whether the game is standard where they have their best chances, or blitz where >>>>it just doesn't lose many games vs GM players, and hardly any any vs IM >>>>players... >>> >>> This is not exactly correct. In 2000+ blitz and bullet games on ICC >>>against crafty clones (Data, Singacrafty, MrsLovett, CraftyCrafty, KillerGrob, >>>Perplexity, Fitter, Brunhilde, Crafty, heatstroke, mundilfari, Mofongo and >>>razzle) I have scored about 7% with about 4% wins by playing anti-computer chess >>>(generally playing to win). My over-the-board strength is probably around 1900 >>>(USCF) a bit below IM. >>> Of course this depends on the definition of essentially zero. >>> James B. Shearer >>>PS: For me Ferret is considerably tougher (1 win, 4 draws in 100+ games). >> >> >>I do not count those programs. Several only play oddball openings. IE Data, >>for the longest, would play _nothing_ but the Evans as white. And would repeat >>until it lost, where BT would then fix the line (he was simply working on >>studying the evans gambit using Data..) Ditto for killergrob, and so forth. > > Actually I think Singacrafty is the weakest of the bunch. The others all >had pretty good ratings as I recall, somtimes better than Crafty itself. Data >never played the Evans gambit against me (as I don't respond e5 to e4). > >>The only programs I can 'control' are Scrappy and Crafty (Scrappy on FICS on my >>quad P6/200, Crafty on ICC or FICS on my quad xeon/400). I don't try to play >>unorthadox openings... I only try for enough 'variety' to avoid playing the >>same line too many times... > > So Scrappy and Crafty are the only "real" crafties? > My score against Crafty is 4 wins and 4 draws in 160 games (2% wins, 4% >overall). Not too good but not zero either. > >>Anyone that would like to play a match vs Crafty is more than welcome. We can >>set it up here, and have at it on either ICC (if you are a member) or on FICS. >>I do _not_ like to play such matches with 'guests' as if either side gets >>disconnected, the game can't be resumed. I will certainly allow Crafty to play >>them unrated however... >> >>Offer open to anyone that wants to try their luck. It only helps me make it >>better... > > Well, I am not sure we are disagreeing. I would expect at least one win >in 100 blitz games (assuming no changes in Crafty as the match progresses). >However you may feel this is essentially 0. > James B. Shearer There are two issues. 1 in 100 is pretty close to zero and probably within acceptable bounds. the main issue is 'crafty' on ICC. Note that it runs 24 hours a day, 7 days a week. But it doesn't take much to figure out that I do lots of other things on this machine, so crafty runs at a "nice 20" level. If you play it during the day, you are _guaranteed_ to play at least one game where it hardly searches at all for several consecutive moves, such as when I am making a new linux kernel, or compiling gcc, or working on various other programs I fiddle with all the time, including crafty itself. It is very likely that out of 30 games in one day, one or two will be like playing crafty on an old 8086, nodes-per-second wise, because of this. That is why I usually encourage anyone to specifically ask to play a match so that I can set things up where this doesn't happen, and you get to play the 'real' crafty without it getting crushed when another compile is fired off... I save _all_ pgn games from both crafty and scrappy... what was your handle when you played and I'll see if I can scrape up the pgn for the losses to see what might have gone wrong, if anything...
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.